Value log-based checkpoint implementation method and system, storage medium and equipment

Abstract

The invention discloses a value log-based checkpoint implementation method and system, a storage medium and equipment, and the method comprises the steps: S1, storing an insertion operation in a memory in a value log implementation process; S2, acquiring a log serial number corresponding to a current check point of a value log system; S3, freezing and storing the memory area of the insertion operation, and changing the attribute of the memory area into read-only; adding the frozen memory area into a to-be-written disk queue, and setting a new writable area for storing insertion operation in the memory; S4, writing to-be-written content into the disk queue into a disk; and S5, writing the log serial number corresponding to the current check point into a check point control file in the disk. According to the method, the check point operation time and the recovery time during downtime can be shortened.

Description

technical field [0001] The invention belongs to the technical field of databases, and in particular relates to a value log-based checkpoint realization method, system, storage medium and equipment. Background technique [0002] Due to the large gap between the speed of the CPU and the speed of the disk, various databases and key-value and other data processing systems are introduced into the data buffer pool, and the data pages of the database are loaded into the buffer pool of the memory. The data update is done in the buffer pool, which means that the data pages are "dirty", that is, the data pages in the buffer pool are newer than the pages in the data files on the disk, and some mechanisms need to be used to refresh the data pages in the buffer plate. [0003] If you start flushing every time a data page is changed, the performance of the database will become very poor. At the same time, if a downtime occurs at this time, the data cannot be recovered. In order to ensu...

AI Technical Summary

Problems solved by technology

[0006] However, because traditional checkpoints have to wait for all "dirty" data pages to be flushed, on the one hand, "dirty" data pages need to occupy a large amount of memory space until the corresponding pages are flushed (corresponding to random write operations); These "dirty" pages maintain a locked state to prevent data from being modified by subsequent update transactions, affecting system operation

Images